Oblique, High-Angle, Sichuan, Chinascec.ess.ucla.edu/~zshen/publ/Areps10Zhang.pdfOn May 12, 2008, a devastating earthquake struck densely populated Sichuan Province, China ... Sichuan

EA38CH14-Zhang ARI 1 February 2010 7:35

R

E V I E W

S

IN

AD V A

NC

E

Oblique, High-Angle,Listric-Reverse Faultingand Associated Developmentof Strain: The WenchuanEarthquake of May 12, 2008,Sichuan, ChinaPei-Zhen Zhang,1 Xue-ze Wen,2 Zheng-Kang Shen,1

and Jiu-hui Chen1

1State Key Laboratory of Earthquake Dynamics, Institute of Geology,China Earthquake Administration, Beijing 100029, China; email: [email protected],[email protected], [email protected] Seismological Bureau, Chengdu 610041, China; email: [email protected]

Annu. Rev. Earth Planet. Sci. 2010. 38:351–80

The Annual Review of Earth and Planetary Sciences isonline at earth.annualreviews.org

This article’s doi:10.1146/annurev-earth-040809-152602

Copyright c© 2010 by Annual Reviews.All rights reserved

0084-6597/10/0530-0351$20.00

Key Words

cascade earthquake rupture, stress accumulation and release, eastern Tibet,Longmen Shan fault, triggered slip

Abstract

The 2008 Wenchuan earthquake occurred on imbricate, oblique, steeplydipping, slowly slipping, listric-reverse faults. Measurements of coseismicslip, the distribution of aftershocks, and fault-plane solution of the main-shock all confirm this style of deformation and indicate cascading earth-quake rupture of multiple segments, each with coseismic slip occurring inthe shallow crust above a depth range of 10 to 12 km. Interactions amongthree geological units—eastern Tibet, the Longmen Shan, and the Sichuanbasin—caused slow strain accumulation in the Longmen Shan so that mea-surable preearthquake slip was minor. Coseismic deformation, however, tookplace mostly within the interseismically locked Longmen Shan fault zone.The earthquake may have initiated from slip on a fault plane dipping 30–40◦

northwest in a depth range from 15 to 20 km and triggered oblique slipon the high-angle faults at depths shallower than 15 km to form the greatWenchuan earthquake.

351


INTRODUCTION

On May 12, 2008, a devastating earthquake struck densely populated Sichuan Province, China(31.0◦N, 103.4◦E) (CENC 2008). The event is named the Wenchuan earthquake as its epicenter islocated in the administrative region of Wenchuan County. More than 80,000 people were killed,more than 370,000 people were injured, and economic losses were estimated at approximately800 billion RMB (nearly $100 billion USD) (CENC 2008). The earthquake generated numerouslandslides that destroyed lifelines, annihilated villages that clung to the steep slopes, and posedsignificant difficulties for rescue efforts. Both the surface-wave magnitude (Ms 8.0) reported by theChina Earthquake Networks Center (CENC 2008) and the moment magnitude (Mw 7.9) givenby different agencies and scientists ( Ji 2008, Wang et al. 2008, Y. Zhang et al. 2008) attest tothis event being a great earthquake. It is China’s most disastrous event since the 1976 Tangshanearthquake, which killed more than 240,000 people.

Although the steep western margin of the Sichuan basin is known to be seismically active, few, ifany, earth scientists anticipated an event of this magnitude to occur there. Two main observationscontributed to this biased view. First, Global Positioning System (GPS) measurements (King et al.1997, Chen et al. 2000, Zhang et al. 2004, Shen et al. 2005, Gan et al. 2007) and active faultingstudies (Burchfiel et al. 1995, 2008; Densmore et al. 2007; Zhou et al. 2007) reveal very slow (<2–3 mm year−1) slip rates across the Longmen Shan fault zone, indicative of relatively modest strainaccumulation and therefore a slowly accumulating seismic hazard. Second, the most devastating,great reverse- or thrust-faulting historic earthquakes commonly rupture gently dipping thrust(not high-angle reverse) faults along which slip occurs rapidly (>50 mm year−1 at the oceanicsubduction zones and >15 mm year−1 along the Himalaya collision zone). For example, the greatearthquakes in Chile (1960), Alaska (1964), and Sumatra (2004) are all associated with geologicallyaveraged slip rates that exceed 50 mm year−1 along gently dipping subduction-zone interfaces(Plafker 1969, Kanamori & Anderson 1975, Ammon et al. 2005, Lay et al. 2005, McCaffrey 2009).In addition, for the earthquakes in Kangra (1905), Bihar Nepal (1934), and Assam (1959), alongthe <15◦ dipping the Main Himalayan Thrust long-term slip rates are 15–20 mm year−1 (Lave& Avouac 2000; Bilham et al. 2001; Kumar et al. 2001, 2006; Avouac 2003; Bollinger et al. 2004;Lave et al. 2005). The 2008 Wenchuan earthquake, however, occurred on a high angle–dippinglistric-reverse fault with a slip rate of less than 2–3 mm year−1 (Densmore et al. 2007, P.Z. Zhanget al. 2008, Zhou et al. 2007). To the best of our knowledge, the 2008 Wenchuan earthquake isthe first with such a large magnitude to have occurred on a slowly slipping listric-reverse faultwithin continental interior during instrumentally recorded earthquake history.

TECTONIC FRAMEWORK OF THE LONGMEN SHAN FAULT ZONE

The Wenchuan earthquake ruptured several strands of the Longmen Shan fault zone, which liesalong the middle segment of the Central Longitudinal Seismic Belt (CLSB) of China, along whichmore than 40 historical earthquakes of magnitude 7 and higher have occurred since the beginningof documented Chinese history (Figure 1). This belt separates the seismically active Tibetan

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 1Regional tectonic map of the Longmen Shan region. (a) Topography, active faults, and earthquakes of the Tibetan Plateau.Earthquakes of magnitude higher than 6 are shown. White dashed polygon outlines approximate the area of the Central LongitudinalSeismic Belt (CLSB). (b) Active tectonic map of the Longmen Shan region [detail area from red-rectangle region in (a)]. GPS velocityvectors are relative to the South China block. Major active tectonic terrains are denoted by their names.

352 Zhang et al.


100˚E 102˚E 104˚E 106˚E 108˚E28˚N

30˚N

32˚N

34˚N

0 5

Elevation (km)

N

Sichuan basin

Eastern Tibet

Ch

ua

nd

ian

blo

ck

CL

SB

Historical earthquakes


M 8.0–8.9

M 7.0–7.9

M 6.0–6.9

10 mm year–1

Western Qinling

a

b

M 8.0–8.9

M 7.0–7.9

M 6.0–6.9

Major active fault

Fault with thrust component

Major active fault

Other active faultSB

Tibetan

plateau

Ordos

block

Tarim block

SouthChinablock

Ch

d

Chengdu

Xianshuihe

Longm

en

Shanfault

fault

Kunlun

fault

Longriba

faultQingchuan

fault

Western Qinlingfault

May 12, 2008 (M 8.0)

May 12, 2008 (M 8.0)

40°N

75°E 85°E 95°E 105°E

35°N

30°N

25°N

www.annualreviews.org • The Wenchuan Earthquake of 2008 353


Plateau from the tectonically stable Ordos block, Sichuan basin, and South China block (Zhanget al. 2003, Zhang 2008). To its west, elevations reach more than 4000 m above sea level, the crustthickens to 50–60 km, and widespread active faulting attests to continued seismotectonic activity(Wang et al. 2007, Yao et al. 2008, L. Xu et al. 2007, Liu et al. 2009). East of this belt, however, theelevation of the Sichuan basin lies only ∼600 m above sea level, the crustal thickness is only 40 to45 km, and the absence of active faulting suggests relatively mild tectonic activity and a relativelylow level of seismicity.

The active Longmen Shan fault zone marks a predominantly convergent boundary with aright-lateral strike-slip component. This fault system was reactivated during late Cenozoic timealong a Mesozoic orogenic belt (Burchfiel et al. 1995, 2008; Kirby et al. 2002, 2008). To the westof the Longmen Shan, eastern Tibet (Songpan–Ganzi Terrain in geological terminology) activelydeforms by both right-lateral shear parallel to and convergence perpendicular to the LongmenShan fault (King et al. 1997, Chen et al. 2000, Zhang et al. 2004, Shen et al. 2005, Gan et al.2007). Tectonic activity in the Sichuan basin, east of the Longmen Shan, has been mild duringlate Cenozoic time. Three principal, subparallel, active faults comprise the northeast-trendingLongmen Shan fault zone (Figure 1). The Yingxiu-Beichuan fault, the major strand, coincideswith the dramatic changes in the steepness of rugged topography (Kirby et al. 2008). The primaryrupture of the 2008 Wenchuan earthquake occurred on this strand (X. Xu et al. 2008). The 2008earthquake also ruptured a 73-km section of the Guanxian-Jiangyou fault, the eastern strand ofthe fault zone along the mountain front. The Wenchuan-Maoxian fault follows the Minjiang rivervalley approximately 30 kilometers northwest of the Yingxiu-Beichuan fault. The 2008 Wenchuanearthquake did not rupture this fault, but numerous landslides and debris flows occurred along it.

Average slip rates for the past ∼10,000 years, however, have been estimated to be quite slow, ap-proximately 0.3–0.6 mm year−1 for reverse faulting and ∼1.0 mm year−1 for right-lateral strike-slipfaulting on the Yingxiu-Beichuan fault, and 0.2 mm year−1 for reverse faulting on the Guanxian-Jiangyou fault (Densmore et al. 2007, Zhou et al. 2007). Ran et al. (2008) reports that the secondterrace above the Minjiang riverbed at the town of Yingxiu has been displaced for 7.6 m includ-ing 2.4 m of coseismic offset during the Wenchuan earthquake. Numerous dates suggest thatthe terrace was abandoned between 3400 and 4000 years ago (Ran et al. 2008). Initiation of thisterrace offset should postdate this abandonment age, and that age gives a minimum value of theslip rate. To remove the contribution of the recent earthquake cycle to the long-term slip rate, wesubtracted the 2.4-m coseismic displacement of the 2008 Wenchuan earthquake from the totalvertical offset of the terrace (Zhang et al. 1988, Li et al. 2009). Thus for the ∼5.2-m, pre-2008 dis-placement of the terrace, the minimum long-term vertical slip rate would be 1.3 to 1.5 mm year−1,which roughly coincides with previous estimates (Densmore et al. 2007, Zhou et al. 2007). Suchgeologically inferred average slip rates appropriate for several thousand years are consistent withGPS estimates of the shortening rate across the Longmen Shan range: <2–3 mm year−1 basedon geodetic measurements made during the past two decades (King et al. 1997; Chen et al. 2000;P.Z. Zhang et al. 2004, 2008; Shen et al. 2005; Gan et al. 2007).

MAINSHOCK AND ITS RUPTURE PROCESS

The China Earthquake Networks Center reported an origin time (14:28:04 Beijing time) and ahypocenter location of the earthquake (31.0◦N, 103.4◦E, focal depth 14.5 km) (CENC 2008).The U.S. Geological Survey gave a similar result. A seismic array (the Western Sichuan SeismicArray, or WSSA) consisting of 297 broadband seismometers was deployed in the western Sichuanregion at the end of 2006 (Liu et al. 2009). The mainshock took place within the northeastern

354 Zhang et al.


quadrant of the array, allowing precise determination of hypocenter locations of the mainshockand aftershock sequence using local records.

Twenty WSSA seismic stations are located within a 75-km radius around the epicenter, and12 of them recorded clear P-wave arrivals on their vertical components. Waveforms of these localstations show small direct waves that rapidly decrease in amplitude as they travel away from theepicenter. These small-amplitude waveforms cannot be seen in records of far-field stations, butthey constrain crucial information about the precise hypocenter location of the mainshock. Theimproved results (Chen et al. 2009) show that initiation of the mainshock occurred at 14 h, 27 min,and 57.10 ± 0.03 s, and that the hypocenter is at 31.0032◦ ± 0.0024◦N, 103.3694◦ ± 0.0025◦E,and 18.7 ± 0.5 km.

Using 18 P-wave first motions and 27 P-wave waveforms from far-field seismic records, Wanget al. (2008) determined the focal mechanism of the mainshock: a seismic moment of 6.5 ×1020, a strike of 229◦, a dip of 32◦ toward the northwest, and a rake of 118◦ with predominantlyunilaterally propagation of the rupture to the northeast. The fault-plane solution indicates mainlythrust slip with a right-lateral strike-slip component, corroborating geological observations ofsurface ruptures associated with this earthquake. Long-period P-waveform inversions performedby different parties show consistent results ( Ji 2008, Nishimura & Yagi 2008, Y. Zhang et al. 2008,Global CMT Project).

Wang et al. (2008) used a finite fault model to determine the temporal and spatial evolutionof the rupture process (e.g., Hartzell & Heaton 1983). They calculated the distribution of slip ontwo listric-thrust faults by combining geological information and focal mechanisms and using acombined inversion of the teleseismic waveforms and GPS coseismic displacements. The inversionshows that rupture started near the town of Yingxiu and propagated along the Yingxiu-Beichuanfault. The calculated slip distribution indicates large slip at depth on the Yingxiu-Beichuan fault,which coincides with field geological observations (X. Xu et al. 2009). Two significant slip patchesare clearly shown in the slip distribution. The southern one, below Hongkou at depths between10 and 15 km, shows a maximum slip between 10 and 12 m. The northern patch, which is betweenBeichuan and Pingtong, shows a slip of ∼12 m at shallow depth. The Tongji-Hanwang segmenton the Guanxi-Jiangyou fault also ruptured for approximately 72 km with 5–6 m slip at depth.Along the Qingchuan segment of the Yingxiu-Beichuan fault, where the rupture does not reachthe surface, 3–4 m slip is also inferred at depth.

To reveal details of the rupture processes associated with the Wenchuan earthquake, Y. Zhanget al. (2009) used a linear-inversion technique, assuming constant rupture velocity and fixed-sourcetime functions of subevents, to image the spatial-temporal variation of the source mechanismbased on 48 worldwide teleseismic waveforms. They show that the entire rupture process canbe approximated as four successive subevents, with each breaking a different section of rupturezone (Y. Zhang et al. 2009). The first subevent characterizes the first 10 s of slip and ruptured∼40 km of the fault with almost pure reverse faulting. It released approximately 7% of the totalseismic moment, equivalent to a Mw = 7.5 event. The second subevent occurred between 10 sand 42 s with largely reverse slip plus a small right-slip component. It ruptured approximately100 km and accounts for 61% of the total seismic moment. The third subevent ruptured a ∼60-kmstretch of the fault with a large strike-slip component and released only 9% of the total seismicmoment. The fourth subevent started 60 s after slip began and continued to the end of rupture(95 s from initiation), releasing 23% of the total seismic moment. Strike slip mostly occurredwith a small reverse-faulting component along its length of approximately 110 km. The overallprocess includes a rupture that starts from the southwestern end as reverse faulting and propagatesunilaterally northeastward for 300 to 340 km with an increasing strike-slip component (Y. Zhanget al. 2009).



SURFACE RUPTURES ASSOCIATED WITH THEWENCHUAN EARTHQUAKE

Postearthquake field investigations indicate that the 2008 Wenchuan earthquake occurred onmultiple imbricate, high-angle (60◦ to 80◦) reverse faults. The rupture zone consists of at least threestrands (X. Xu et al. 2008, 2009): the NE-trending Yingxiu-Shuiguan rupture on the NE-trendingYingxiu-Beichuan fault, the Tongji-Hanwang rupture on the NE-trending Guanxian-Jiangyoufault, and the short, NW-trending, arcuate Xiaoyudong rupture that apparently intersects theLongmen Shan fault (Figure 2).

The Yingxiu-Shuiguan Rupture

For the Yingxiu-Beichuan fault, the primary structure responsible for the Wenchuan earthquake,the rupture started near the town of Yingxiu (30.986◦N, 103.364◦E); crossed towns and villagessuch as Hongkou, Gaochuan, Beichuan, Pingtong, and Nanba; and terminated east of Shuiguan(32.2862◦N, 104.9515◦E) with a length of 240 km (Figure 2). Discontinuous ruptures such asfissures and cracks could be found for an additional 20 km to the northeast, but no displacementcould be measured along this northeastward continuation. For another ∼60 km along the Yingxiu-Beichuan fault beyond the northeastern end of the surface rupture and fissuring zone, frequentaftershocks delineated an additional rupture segment that did not reach the surface (Figure 2).Also, southwest of the southwestern end of the mapped surface rupture, aftershocks extended foranother ∼20 km, which makes the total length of the aftershock zone—and presumably of therupture itself—340 km (Figure 2).

Primarily reverse faulting characterizes the southern fault segment from Yingxiu toXiaoyudong. At Yingxiu, the surface rupture cuts through the prosperous tourist town and causedsignificant damage and many casualties. A paved road on the terrace of the Minjiang river has beenoffset 2.2 ± 0.3 m vertically, but no horizontal offset has been observed (Figure 3a). Northeast-ward approximately 10 km near Hongkou, the rupture displaced roads, farming fields, and farmers’houses to form prominent fault scarps. The vertical offsets increase to 4 to 5 m, and right-lateraldisplacements of 1 to 3 m were also measured. At one place (31.10472◦N, 103.62225◦E), the largestvertical offset of 6.2 ± 0.5 m of this segment was measured (X. Xu et al. 2009). A seismogenicscarp at Bajiao (31.14522◦N, 103.69189◦E) dips 76◦ northwestward, and slickenside striations onthe fault surface indicate that the fault slipped at 75 to 82◦ subvertically (Figure 3b). This testifiesto a prominent component of reverse faulting.

North of the Xiaoyudong fault to Gaochuan, surface ruptures are present in the high mountainsof Longmen Shan. Numerous landslides and earthquake lakes prohibited access to most of theplaces, so only four localities have been studied. The steep topography does not preserve surfaceruptures, and only in flat valleys can ruptures be observed. The vertical offsets in this segmentaverage 3 to 4 m with an average of ∼2 m of right-lateral offset. For example, the surface rupturedisplaced a paved road 3.7 ± 0.3 m vertically and 0.55 ± 0.2 m right-laterally near the town ofQingping (31.34071◦N, 104.06715◦E) (Figure 3c).

From north of Gaochuan to Shuiguan, the coseismic rupture occurred on a single fault, alongwhich both maximum vertical and horizontal offsets occurred (Figure 2). At Beichuan and to itsnortheast, a subvertically dipping surface rupture offset ridges and gullies sharply in places to formscarps whose heights vary from 1–2 m up to ∼9 m (Figures 2b and 3d ). The maximum verticaloffset is observed northeast of the town of Beichuan where the surface rupture cuts througha farmer’s yard (Figure 4a,b). Houses adjacent to the three-story building and to a flat, 8 m–wide yard were destroyed, as the rupture offset passed beneath both those houses and the yard

356 Zhang et al.


Pingwu

Mianyang

Pingtong

Nanba

Shuiguan

Beichuan

Sangzao

Hanwang

Gaochuan

Qingping

BailuTongji

Wenchuan

Maoxian

HongkouYingxiu Xiaoyudong rupture

Yingxiu

- Shuig

uan ruptu

re

32.5°N32.5°N

31.5°N 31.5°N

104°E

104°E 105°E

105°E

Tongji - H

anwang ru

pture

Pengguan massif

Wenchuan-Maoxian fa

ult

Min

jiang

faul

t

Hu

ya

fa

ult

Guanxian-Jiangyou fault

Qingchuan fault

Hongkou

segment

Hanwang

segment

Beichuan segment

Qingchuan

segment

The mainshock

a

b

Jiaoziding massif

c

0 20 40 60 80 100 120 140 160 180 200 2200

2

4

6

8

10

Distance along the Yingxiu-Shuiguan rupture (km)

Co

seis

mic

off

sets

(m

)

Yingxiu

Hongkou

Gaochuan

Beichuan

Pingtong

Coseismic vertical displacement Right-lateral coseismic displacement

Pingxi

40 60 80 100 1200

1

2

3

4

5

6

Co

seis

mic

off

sets

(m

)

Distance along the Tongji-Hanwang rupture (km)

Bailu

Hanwang

Bajiao

Surface rupture

Active fault

Town or village

Approximate location of segmentation boundary of the surface rupture zone

M 8.0–8.9

M 7.0–7.9

M 6.0–6.9

M 5.0–5.9


Precambrian massif

Figure 2Surface ruptures and coseismic displacements associated with the 2008 Wenchuan earthquake. (a) Active faults and surface ruptures.(b), (c) Coseismic slip distributions along the Yingxiu-Beichuan fault and the Guanxian-Jiangyou fault, respectively.



a b

c d

e f

Figure 3Photos of surface ruptures associated with the Wenchuan earthquake. (a) Fault scarp formed during the earthquake at the town ofYingxiu near the southern end of the rupture zone. View is to the northwest. The road and the southern bank of the Minjiang riverhave been offset for 2.2 ± 0.2 m. A person standing at the foot of the scarp provides scale. (b) Subvertical fault scarp at Bajiao villagenear Hongkou town. Surface rupture shown in the photo dips 76◦ northwestward, and slickenside striations on the fault surface indicatethat fault slipped at an angle of 75–82◦ subvertically. View is to the northeast. (c) Fault scarp near the town of Qingping, where a pavedroad has been offset 3.7 ± 0.3 m vertically with a minor right-lateral slip of 0.55 ± 0.2 m. View is to northwest. (d ) Step-dipping faultscarp near the town of Beichuan. View is to the southwest. The scarp height is 1.8 ± 0.4 m. (e) Schoolyard at the town of Bailu, offsetby the Tongji-Hanwang rupture. The yard was flat and the two buildings were the same height before the earthquake. The verticaldisplacement was measured to be 1.8 ± 0.3 m. View is to the northeast. ( f ) Vertical and left-lateral slips along the northwest-trendingXiaoyudong fault. A paved road is offset with left-lateral offset 2.3 ± 0.3 m and vertical offset of 1.5 ± 0.4 m.

358 Zhang et al.


c

4.9±0.4 m

d

644644

643643

641

643

641

644

641641

8.6±0.5 m

a

UncollapsedUncollapsedthree storythree storybuildingbuilding

Concrete floorConcrete floorof the yardof the yard

UncollapsedUncollapsedtwo storytwo storybuilding building

Uncollapsedthree-storybuilding

Concrete floorof the yard

Uncollapsedtwo-storybuilding

PreservedPreservedbasement ofbasement ofa collapseda collapsedbuildingbuilding

Preserved concretePreserved concretefloor of the yardfloor of the yard

Destroyed concreteDestroyed concretefloor of the yardfloor of the yard

67

5

67

5

68

0

685

68

5

b

0 5 10 m

S

Preservedbasement ofa collapsedbuilding

Preserved concretefloor of the yard

Destroyed concretefloor of the yard

Figure 4The maximum vertical and horizontal coseismic displacements. (a) A farmer’s yard offset by the rupture. The inset photo shows theyard before the earthquake. The houses opposite the three-story building were destroyed as the rupture offset the flat and 8 m–wideyard. A portion of the yard has been preserved on the hanging wall (the concrete surface where survey equipment is located) after theearthquake. (b) Detailed total station map of the offset shown in (a). The separation of the preserved concrete yard from its counterpartin front of the building yields 8.6 ± 0.5 m of vertical offset. (c) The maximum right-lateral offset near the town of Pingtong. A smalltrail within a farming field has been offset for 4.0 ± 0.3 m right-laterally. (d ) Total station map of the offset shown in (c).

(Ran et al. 2010). The inset photo shows the yard and three-story building before the earthquake(Figure 4a). A portion of the yard has been preserved on the hanging wall (the concrete surfacewhere survey equipment is shown in Figure 4a). The separation of the preserved concrete yardfrom its counterpart in front of the building yields 8.6 ± 0.5 m of vertical offset (Ran et al. 2010).The maximum right-lateral offset is located near the town of Pingtong, where a small trail withina farming field has been offset 4.9 ± 0.3 m right-laterally (Figure 4c,d ).

Further northeastward, the right-lateral strike-slip component increases and becomes domi-nant at Pingtong and further northeast. For example, subvertically dipping surface ruptures areprominent at Pingxi ∼4 km from the northern end of the surface rupture zone, where the verticaloffset is 1.75 ± 0.25 m and the right-lateral slip reaches 3.0 ± 0.5 m.



The Tongji-Hanwang Rupture

Approximately 10 to 12 km southeast of and parallel to the Yingxiu-Shuiguan rupture, the Tongji-Hanwang rupture follows a section of the Guanxian-Jiangyou fault for 72 km (Figure 2). Therupture also dips northwestward with predominantly reverse motion (X. Xu et al. 2009). Therupture starts near the town of Tongji where scarp heights vary from 1 to 2 m without ob-vious horizontal offset. The vertical offset increases northeastward. At the town of Bailu, therupture cuts through a schoolyard between two school buildings to form a prominent fault scarp(Figure 3e). The measured scarp height is 1.8 ± 0.3 m with no horizontal offset. The nearestdistance of the fault scarp to one of the school buildings is only 3 m. Fortunately, the buildingsdid not collapse and none of the more than 300 elementary-school students were even injured.Further northeast, X. Xu et al. (2009) reported a vertical offset of 3.5 m, which is the largestalong this segment. The vertical offset remains approximately 1 m near the intensively damagedindustrious town of Hanwang. The rupture terminates near the town of Sangzao with the scarpheight decreasing to 0.2 to 0.3 m.

The Xiaoyudong Rupture

Northwest of the southwestern termination of the Tongji-Hanwang rupture is a 6 km–long,arcuate surface rupture named the Xiaoyudong rupture zone (Figure 2). This remarkable ruptureis characterized by its change in strike from northeast in the southwestern part to north-northwestin the northeastern part. The north-northwest-trending section has a predominant left-slip com-ponent, whereas the north-northeast-striking section shows mostly reverse faulting. Figure 3f

shows that the Xiaoyudong rupture offsets a paved road with a left-lateral offset of 2.3 ± 0.3 mand a vertical offset of 1.5 ± 0.4 m. The Xiaoyudong rupture apparently offsets the Yingxiu-Beichuan rupture zone left-laterally by approximately 4 km (Figure 2), and it aligns with thenorthwest-trending (and protruding) aftershock zone west of the town of Yingxiu (see below).

Geometric Segmentation and Slip Distribution

To summarize, the Wenchuan earthquake occurred by slip on multiple imbricate, high-angle (60–80◦) reverse faults with right-lateral components of slip. The surface rupture is 240 km in lengthwith a maximum vertical offset of ∼9.0 m and a right-slip component of ∼4.9 m. It is also evidentthat vertical components dominate the coseismic deformation except near the northeastern endwhere right-lateral strike slip becomes predominant (Figure 2b,c). The entire rupture zone canbe divided into four segments, each of which has a different geometric pattern and kinematicbehavior (Figure 2a). The geometric complexity as manifested by multisegmented and imbricatesurface ruptures indicates structural discontinuities or barriers along the seismogenic fault thatrequire complex rupture processes to overcome them (Sibson 1986, 1989; Zhang et al. 1991; Yeatset al. 1997). The four-segment nature coincides with teleseismic waveform-inversion results oftemporal and spatial variations of the earthquake rupture ( Ji 2008, Nishimura & Yagi 2008, Wanget al. 2008, Y. Zhang et al. 2008).

The envelopes of coseismic slip distribution along the ruptures corroborate the segmentedgeometric pattern of the zone (Figure 2). Each segment reveals a different pattern of slip distri-bution. The Hongkou segment has a single peak near its middle portion. The Hanwang segmentinvolves slip partitioning between two parallel strands with a total average offset in the range of6–8 m (Figure 2). The Beichuan segment has a peak vertical offset in its southern portion and apeak horizontal offset near its middle. The Qingchuan segment has no surface displacement.

360 Zhang et al.


105°E

Maoxian

Lixian

Songpan

PingwuPingwu

BeichuanBeichuan

ChengduChengdu

DujiangyanDujiangyan

WenchuanWenchuan

Hongkou segment

Hanwang segment

Beichuan segment

Qingchuan segment

Qingchuan segment

Qingchuan segment

Pingwu

Qingchuan

Guangyuan

Jiange

Beichuan

Mianyang

Deyang

Chengdu

Dujiangyan

Wenchuan

Bikou massifBikou massifBikou massif

Baoxing massifBaoxing massifBaoxing massif

Pengguan massif

103°E

103°E

104°E 106°E

104°E 105°E 106°E

30.5°N

31.5°N

32.5°N

30.5°N

31.5°N

32.5°N

104˚E

31˚N

30˚N

32˚N

33˚N

34˚N

0 50km

105˚E102˚E 103˚E 106˚E

Aftershocks

M 8.0–8.9M 7.0–7.9M 6.0–6.9M 5.0–5.9M 4.0–4.9M 3.0–3.9M 2.0–2.9

Active fault

Precambrian massif

Town or village

Surface rupture

Beichuan

Anxian

Lixian

WenchuanHanwang

Qingchuan

Seismic stations

Figure 5Relocated aftershock distribution of the Wenchuan earthquake. Boxes indicate aftershocks of each segment used to construct profiles inFigure 6. The lower-right inset shows distribution of seismic stations used for the aftershock relocations. Triangles are stationsaffiliated with the Western Sichuan Seismic Array (WSSA) deployed since 2007; diamonds are from the Sichuan Seismic Network andsome temporary stations deployed after the mainshock.

AFTERSHOCK DISTRIBUTIONS AND THE STRUCTURAL PATTERNOF THE LONGMEN SHAN FAULT ZONE

Precise relocations of the aftershocks can illuminate subsurface structures of ruptures (e.g., Carenaet al. 2002, Hauksson & Shearer 2005). As mentioned above, the mainshock occurred within thenortheastern quadrant of the WSSA. The array also spans 2/3 of the aftershock zone. Togetherwith the Sichuan Seismic Network and other seismic stations deployed immediately after theearthquake, the stations shown in Figure 5 allow precise locations of the aftershocks. To avoidcomplications arising from other wave trains, Chen et al. (2009) used only stations within 100 kmof the epicenter for relocations of each event and selected a total of 3920 events for relocationusing the double-difference software hypoDD of Waldhauser & Ellsworth (2000). Each of theseevents was recorded by at least eight stations with a high signal-to-noise ratio. Finally, 3622 eventswere relocated with uncertainties estimated to be ± 0.85 km and ± 1.75 km for horizontal andvertical coordinates, respectively (Chen et al. 2009).

Unlike the 1999 Chi-chi earthquake in Taiwan, in which most of the aftershocks lie along agently dipping zone for ∼100 km in the hanging wall (Carena et al. 2002, Chang et al. 2007),



most of the aftershocks are narrowly distributed within a zone 15–30 km wide in the hanging wallof the Yingxiu-Beichuan fault (Figure 5). The relatively narrow distribution offers no evidenceof thin-skin thrust faulting. The overall pattern of the hypocenter distribution coincides with thesegmented nature derived from surface ruptures (Figure 5), indicating how fault structure providesa geometric control on the aftershock distribution. Aftershocks in the Hongkou segment clusterwithin the main body of the Precambrian Pengguan massif and suggest the possibility of a high levelof residual or redistributed stress after the mainshock. A northwest-trending aftershock zone cutsthrough the massif. In the Hanwang segment, most of the aftershocks occurred in the hangingwall of the Yingxiu-Beichuan rupture, and a minor portion occurred within the hanging wallof the Tongji-Hanwang rupture. The boundary between the Hanwang and Beichuan segmentscorresponds spatially to a broad region of aftershocks (Figure 5). Aftershocks along the Beichuansegment are mostly confined to the hanging wall of the earthquake rupture. The absence ofaftershocks in the footwall along this segment implies that the rupture of the Guanxian-Jiangyoufault does not extend to this segment. In the Qingchuan segment, aftershocks occurred on bothsides of—but close to—the northern extension of the Yingxiu-Beichuan fault (Figure 5). Thispattern is similar to that along typical strike-slip earthquakes, such as the Loma Prieta (1989),Landers (1992), and Parkfield (2004) earthquakes in California (Sieh et al. 1993, Dietz & Ellsworth1997, Thurber et al. 2006).

Most of the relocated aftershock hypocenters occurred between depths of 10 and 22 km (Chenet al. 2009). This pronounced depth distribution suggests that the elastically deformed brittleportion of the crust (schizosphere) may be as deep as 22 km, below which the brittle-ductiletransition lies. It also suggests that significant coseismic slip takes place mainly at shallow depthsabove 10 km, so that previously accumulated stress was released during the mainshock. Studiesof aftershock patterns and mainshock faulting indicate that the spatial distribution of aftershocksreflects either a continuation of slip in the regions with relatively small slip during the mainshockor the activation of subsidiary faults within the volume surrounding the boundaries of mainshockrupture (Ouyed et al. 1983, Mendoza & Hartzell 1988, Hatzfeld et al. 1997, Chang et al. 2007).In other words, aftershocks do not occur where the mainshock slip is large; instead, they occurin regions where the stress level is high following the mainshock. This relation may be used todelineate coseismic faults at depth. For strike-slip faults, the orientation of the aftershock clusteringcommonly delineates the seismogenic fault (Sieh et al. 1993, Dietz & Ellsworth 1997, Thurberet al. 2006). For reverse or thrust faults, the aftershocks mainly occur in the hanging walls withhigh residual stress, and the seismogenic fault is commonly devoid of aftershocks.

Figure 6 shows profiles of aftershock hypocenters across different segments of the Wenchuanearthquake rupture zone that illuminate subsurface seismogenic structures. Most of the aftershocksacross the Hongkou segment lie in the depth range of 10 to 20 km within the imbricate structures,and only a few are sparsely located shallower than 10 km (Figure 6a). The basal envelope ofaftershocks may delineate the base of the schizosphere or brittle-ductile transition zone belowwhich ductile shear accommodates strain. The mainshock took place initially on a ramp dippingnorthwestward at 30–40◦ in the depth range of 15 to 20 km, and at shallower depths the faultsteepens to ∼70◦ to form prominent surface ruptures with significant displacements along theYingxiu-Shuiguan rupture. Eastward the earthquake broke another ramp splay to form the Tongji-Hanwang surface ruptures. Farther east, there may have been some minor slip on the range-frontfault, but it did not reach the surface (Figure 6a). Aftershocks on the Hanwang segment lie mainlyin the depth range of 14 to 22 km between the Yingxiu-Shuiguan and Tongji-Hanwang rupture(Figure 6b). Almost all the aftershocks along the Beichuan segment occurred in the depth range of8 to 18 km, northwest of the Yingxiu-Shuiguan rupture and in the hanging wall (Figure 6c). The

362 Zhang et al.


Interpreted fault

Fault at surfaceHypocenter locationof aftershock

Mainshock location (approximate)

Inferred fault

Distance across the Longmen Shan fault zone (km)

De

pth

be

low

su

rfa

ce (

km

)

WMF YBF GJFb HanwangHanwang

c BeichuanBeichuan

d QingchuanQingchuan

b Hanwang

30

20

10

0

De

pth

be

low

su

rfa

ce (

km

)

WMF YBF GJFa Hongkou

30

20

10

0

De

pth

be

low

su

rfa

ce (

km

)

YBF GJFc Beichuan

30

20

10

0WMF

De

pth

be

low

su

rfa

ce (

km

)

0 10 20 30 40 50 60 70 80 90

d Qingchuan

30

20

10

0QCF YBF GJF

Figure 6Distribution ofrelocated aftershocksacross the Hongkou(a), Hanwang (b),Beichuan (c), andQingchuan (d )segments. Locations ofthe profiles are shownin Figure 5. Solid linesare interpreted faultsbased on geologicaland seismologicalinformation. Dashedlines are inferredfaults. Abbreviations:WMF, Wenchuan-Maoxian fault; YBF,Yingxiu-Beichuanfault; GJF, Guanxian-Jiangyou fault; QCF,Qingchuan fault.



inferred fault also shows a steep dip of ∼70◦ above ∼15-km depth and decreases to ∼40◦ beforemerging to a subhorizontal brittle-ductile transition zone at ∼20 km. In the Qingchuan segment,along which the rupture did not reach the surface, aftershocks cluster around the northeasternextension of the Yingxiu-Shuiguan rupture zone in the depth range of 8 to 22 km along theYingxiu-Beichuan fault. Concurrent with the prominently strike-slip component inferred fromseismological and geodetic observations (Wang et al. 2008, Y. Zhang et al. 2008, Shen et al. 2009),the cross section (Figure 6d ) shows the pattern associated with steeply dipping strike-slip faults(Sieh et al. 1993, Dietz & Ellsworth 1997, Hauksson et al. 2002, Hauksson & Shearer 2005,Thurber et al. 2006).

To summarize, the aftershock patterns reveal three important features. First, most aftershockhypocenters in the depth range of 10 to 22 km suggest that coseismic slip of the mainshock releasedmost of the seismogenic stress buildup in the upper 10 km of depth. Second, both map and cross-section views of the aftershock distribution demonstrate a complex fault system that suggests anearthquake rupture cascading along multiple segments. Third, the geometric pattern of aftershockdistribution appears to favor the high-angle, listric-reverse faults that root into a subhorizontalbrittle-ductile transition below 20–22 km of depth.

Fault-plane solutions for 32 aftershocks of magnitude 5 and higher show a dominance ofreverse faulting along the major part of the rupture zone, the Hanwang and Beichuan segments(Zheng et al. 2009). Fault planes of most of these events trend parallel to the surface rupture zonealong the main strand, and dip 30–80◦ northwestward with focal depths ranging from 11 to 19 km.Horizontal projections of the P-axes of the fault-plane solutions show a consistent west-northwesttrend, subperpendicular to the strike of the Yingxiu-Beichuan fault (Zheng et al. 2009).

Six of the seven fault-plane solutions of major aftershocks northeast of the surface ruptureshow strike-slip faulting; one of these nodal planes strikes parallel to the Yingxiu-Beichuan faultzone. This corroborates the increasing strike-slip displacement along the surface rupture towardthe northeast, obtained through teleseismic waveform inversion of the mainshock ( Ji 2008, Wanget al. 2008, Y. Zhang et al. 2009) and geodetic measurements of coseismic movement. The 15fault-plane solutions near the southern end of the rupture also reveal reverse, strike-slip, and evennormal faulting. We interpret the variety of styles and orientations of faulting to be the result ofcomplexities in stress near the tip of a propagating rupture.

LARGE-SCALE COSEISMIC DISPLACEMENTSFROM GEODETIC OBSERVATIONS

The prominent feature of the displacement field measured with GPS (Working Group 2008) ishorizontal shortening or convergence across the seismogenic Yingxiu-Beichuan fault (Figure 7a).Northwest of the surface rupture, all stations moved eastward and southeastward in a frame of

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 7Coseismic displacements measured by GPS and leveling. (a) Coseismic horizontal displacements measured by GPS in a frame ofreference defined by their preearthquake locations (Working Group 2008). Arrows of different colors represent different magnitudes ofGPS velocity. Thick black lines are surface ruptures associated with the 2008 Wenchuan earthquake. (b) Coseismic horizontaldisplacement profile across the surface rupture zone. Blue squares are components normal to fault, representing coseismic convergence.Green squares are components parallel to fault, representing coseismic right-lateral offsets. (c) Coseismic vertical displacementsmeasured by leveling. The last preearthquake leveling was 1997. The uncertainty and the accumulating slip from 1997 to 2008 areinsignificant compared with coseismic displacements of 4.7 to 0.5 m. The coseismic displacement scale is logarithmic.

364 Zhang et al.


Surface rupture associated with 2008 Wenchuan earthquake

Other active fault Town or villageGPS station

100˚E 101˚E 102˚E 103˚E 104˚E 105˚E 106˚E 107˚E 108˚E 109˚E 110˚E

100˚E 101˚E 102˚E 103˚E 104˚E 105˚E 106˚E 107˚E 108˚E 109˚E 110˚E

29˚N

30˚N

31˚N

32˚N

33˚N

34˚N

29˚N

32˚N

33˚N

34˚N

100 mm

400 mm

Beichuan Bazhong

Aba

SongpanRangtang

Qingchuan

Kangding

Zigong

Ziyang

Guangan

Dazhou

Chongqing

Lichuan

Ankang

Xian

Wanyuan

Chengdu

Confidence 50%

a

Convergence

104˚E

4.7 m

Vertical displacement (m)0 1 2 3 4 5

105˚E

31˚N

32˚N

Mianzhu

Anxian

Beichuan

Jiangyou

Maoxian

Pingwu

c

0 100 200 300 400 500 600

–2

–1

0

1b

Lon

gm

en

Sh

an

fa

ult

Co

seis

mic

dis

pla

cem

en

ts (

m)

Distance across the Longmen Shan fault NW to SE (km)

Right-lateral slip

Coseismic vertical displacement

Town or villageBase station

Maximum coseismicvertical displacement

Earthquake surface rupture

Inferred surface rupture

Other active fault

5000

0

Ele

va

tio

n (

m)



reference defined by their preearthquake locations, whereas stations located southeast of the rup-ture moved westward and northwestward. The observed maximum horizontal coseismic displace-ment is ∼2.5 m west-northwestward at Station H035, which is only ∼2 km from the fault on thefootwall near the town of Beichuan (Figure 7a). Unfortunately, there are no stations close to thefault on the hanging wall, which would have moved farther than that 2.5 m at the same distancefrom the fault trace but in the footwall. In the far field, the amount of displacement seems to decayat different rates on the two sides of the fault, with the hanging-wall displacement decaying moreslowly. For example, Station H046, located ∼120 km west of the fault on the hanging wall, moved267 mm toward the fault, but the component perpendicular to the fault at Station JYAN, alsolocated ∼120 km east of the fault on the footwall, moved only 82 mm (Figure 7a). This differencemay reflect modest slip at depth within or below the brittle-ductile zone northwest of the rupture.

The right-lateral component of displacement associated with the Wenchuan earthquake isalso apparent in the GPS displacement field (Figure 7b). As is apparent from the displacementssoutheast of the fault, there is almost no strike-slip component along the southwestern segment ofsurface rupture; control points on opposite sides of the rupture moved toward one another. Likesurface faulting measurements that indicate right-lateral slip of ∼2–3 m along the central segmentand 3–4 m along the northern segment (Figure 2b,c), apparent components of right-lateral shearin the GPS field also increase to the northeast.

A leveling line across the northern part of the surface rupture zone was resurveyed after theearthquake (the last survey before the earthquake took place in 1997). The largest vertical dis-placement on the hanging wall is measured to be 4.7 m upward, at a site only 168 m from thefault in the town of Beichuan (Figure 6c). The downward movement is 0.5 m at the maximumand decreases rapidly away from the fault on the footwall.

Although the coseismic displacement field can be roughly modeled by a simple elastic dislo-cation consisting of slip on a fault plane that dips at 70◦ in the upper 8 km, at 60◦ between 8to 15 km, and at 40◦ between 15 to 20 km, the large misfit suggests that the deformation is toocomplex to be described by such a simple dislocation model. This problem partly arises from thesparse GPS control points, and especially from the lack of stations on the hanging wall near thefault. Employing a Newtonian nonlinear inversion method, Shen et al. (2009) used 362 GPS and9110 InSAR control points to solve simultaneously for the fault geometry and slip distribution onsegments of faults. The modeled rupture peaks at two places at shallow depth, one near Yingxiuand the other near Beichuan (Figure 7). Obviously, this spatial distribution of slip corroboratesfield observations in general (Figure 2).

The result of inverting the geodetic data again shows a complex fault geometry (Figure 8). Tofit the GPS and InSAR data, the Yingxiu-Beichuan fault must dip to the northwest at a moderateaverage dip angle of ∼43◦ near the southwest end, become steeper northeastward along the strike,reach a dip of ∼50◦ at Nanba, and increase progressively until the fault plane becomes vertical

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 8Geodetic inversion result: rupture geometry and coseismic slip distribution on fault (from Shen et al. 2009). The fault planes are viewedfrom southwest, at a 45◦ elevation angle in all diagrams. (a) Fault geometry. Fault-dip angle is assumed constant along dip and varieslinearly along strike. (b) Coseismic slip distribution from inversion of GPS and InSAR data. The Guanxian-Jiangyou fault (GJF) isplotted away from its actual location. Black arrows show the slip vectors on the fault patches, whose amplitudes are denoted by thecolors of the patches. Red lines are the mapped traces of surface breaks from X. Xu et al. (2009). The brown columns show the densityof aftershocks along the fault, which occurred within 50 km of normal distance to the fault patches at the surface. B1–B11 and G1–G3are fault segments on the Yingxiu-Beichuan fault and Guanxian-Jiangyou fault, respectively.

366 Zhang et al.


Mapped traces of surface breaks

Density of aftershock

103.0

103.5

104.0

104.5

105.0

105.5

30.5

31.0

31.5

32.0

32.5

0

20

40

Latitude (°

N)

Longitude (°E)

De

pth

(k

m)

De

pth

(km

)

1 2 3 4 5 6 7

Slip (m)

30.5

31.0

31.5

32.0

32.5

0

20

40

B1B1

B11B11B10B10B9B9B8B8

B7B7B6B6B5B5B4B4B3B3

B2B2B1

B11B10B9B8

B7B6B5B4B3

B2

G3

GJF

G2G1

Nanba

Beichuan

N

b

Nanba

Beichuan

Yingxiu

N

56°

96°

50°

43°

49°

44°

43°

0

2040

103.0

103.5

104.0

104.5

105.0

105.5

106.0

30.531.0

31.532.0

32.533.0

49°

28°

GJF

aQingchuan

Lon

git

ud

e (°

E)

Latitude (°N)

Qingchuan

Yingxiu



at the northeast end of the rupture (Shen et al. 2009). Thus, in accordance with the geologicalobservations and seismological inferences, high-angle, oblique reverse faulting is required to pro-duce the geodetically observed coseismic displacement field. The segmented nature of geodeticallymodeled ruptures also coincides with geological segmentation along the fault zone (Figures 2 and5) and attests to the fact that the Wenchuan earthquake broke through several high-slip junctionsthat connect major fault segments in a cascading rupture (Figure 8). These connecting structuresmay represent barriers that rarely fail, and do so only after high stress has accumulated after mul-tiple rounds of smaller events have broken the adjoining individual segments (Shen et al. 2009).Such a cascading rupture scenario helps explain why Yingxiu, Beichuan, and Nanba experiencedthe highest shaking intensity of XI on the China Seismic Intensity Scale and suffered the greatestdamage among the towns located along the fault zone.

IMBRICATE, LISTRIC, HIGH-ANGLE, OBLIQUE-SLIP FAULTINGAND SEISMOGENIC STRUCTURES

Despite abundant seismic reflection data available in the Sichuan basin from the petroleum ex-ploration, the deep structure of the Longmen Shan fault zone responsible for the Wenchuanearthquake generation is still not well known. The steep topography across the Longmen Shanprevents acquisition of high-resolution seismic data. The upper crustal structure of the LongmenShan fault zone thus has been inferred from surface geology, a few shallow-level (less than 10 km–deep) seismic reflection profiles, and other geophysical data (Burchfiel et al. 2008, Zhu 2008,Hubbard & Shaw 2009). In the following we provide further evidence showing high-angle, listric-reverse faulting during the Wenchuan earthquake to shed some light on deep structure and pro-cesses that produced and supported the Longmen Shan Mountains.

Surface exposures of the earthquake rupture along the Yingxiu-Beichuan fault show dip anglesof 70–80◦ to the northwest (Figure 3b,d, for example) and 40–60◦ to the northwest along theGuanxian-Jiangyou fault (H. Li et al. 2008; Liu et al. 2008; X. Xu et al. 2008, 2009; Z. Xu et al.2008). Trench excavations across the ruptures also reveal high-angle reverse faulting near thesurface (Ran et al. 2008). For example, a 4 m–deep trench at Bajiao (31.14522◦N, 103.69189◦E)confirms a 76◦ dipping fault plane shown by Figure 3b along the Yingxiu-Beichuan fault; a 10 m–deep trench at Bailu across the Guanxian-Jiangyou fault shows a 47◦ dipping fault plane. Althoughthe general consensus is that surface fault scarps are notoriously poor indicators of subsurfacefault dip, they nevertheless serve as counterexamples to support high-angle rupture at least at thesurface.

Z. Xu (2009) has led a project involving deep drilling of the Wenchuan earthquake rupture.Test drilling was conducted at Bajiao near the outcrop of the surface rupture with clear slickensidestriations (Figure 3b). The drill hole reaches the earthquake fault at a depth of 650 m as predictedby assuming a 70◦ dip of the fault plane (Z. Xu 2009) and therefore testifies to high-angle faultingextending to a depth of at least 650 m.

Fault-zone trapped waves are the wave trains that propagate within the fault zone and thatfollow the S waves on seismograms, usually with large amplitudes and low frequency (2–5 Hz)(Li & Leary 1990, Li et al. 1994). Fault-zone trapped waves have provided a powerful tool forrevealing the subsurface geometry and physical properties of seismogenic fault zones and havebeen successfully exploited in studies of seismogenic faults of the Landers (1992), Hector Mine(1999), and Parkfield (2004) earthquakes (Li et al. 1994, 2002, 2006) as well as the Kunlun (2001)earthquake in the northern Tibetan Plateau (S.L. Li et al. 2005, Wang et al. 2009). Shortly afterthe Wenchuan earthquake, four densely instrumented, temporary seismic arrays were deployed

368 Zhang et al.


a b

ct/s

L3-2008-07-21-15:09 event, ML1.2, 16 records, N1, 0.5-3.0 Hz

2015 25 30 350

2

4

6

8

10

12

14

16

Dip angle 60° Dip angle 70° Dip angle 80°

y

x

z

StationFault zone

17 16 15 14 13 5678910

1112

4 3 2 1

8642086420 86420

Depth (Vp, Vs)(2.08, 1.15) (4.38, 2.50)

200 m

(6.00, 3.42)

(6.50, 3.75)

(2.85, 1.60)

(3.10, 1.75)

(Vp, Vs)

3 km

6 km

10 km

Source70°

Figure 9Fault-zone trapped waves across the Yingxiu-Beichuan fault zone (from S.L. Li et al. 2009). (a) Diagram summarizing parameters andproperties of the fault zone. (b) Filtered seismic waveforms. The wave trains circled by a red ellipsoid are the observed fault-zonetrapped waves. (c) Synthetic seismograms of the fault-zone trapped waves for dip angles of 60, 70, and 80◦.

across the Yingxiu-Beichuan and the Guanxian-Jiangyou faults to record fault-zone guided waves(S.L. Li et al. 2009). Through waveform analysis and synthetic seismogram calculation, S.L. Liet al. (2009) found that the seismic velocity inside the fault zone is approximately one half of thatof the host rocks. The widths of the rupture zone are approximately 200–230 m and 170–200 macross the southern and northern portions, respectively. To produce the recorded fault-zonetrapped waves along each array, the southern portion (near Hongkou) of the fault zone must dip∼70◦ northwestward and the northern portion (near Pingtong) should dip even more steeply at∼80◦ northwestward (Figure 9). Fault-zone trapped waves can detect structures only above theaftershock depth, which is approximately 10 km. Thus, the Yingxiu-Beichuan fault apparently dipsat 70 to 80◦ northwestward to a depth of 10 km.

Geological studies (Burchfiel et al. 1995, 2008; Chen & Wilson 1996; Wang & Meng 2008)and limited seismic reflection profiles from petroleum exploration (Chen et al. 2005, Jia et al.



2006, Hubbard & Shaw 2009) show that the Mesozoic to early Cenozoic structures of theLongmen Shan are characterized by a low-angle, imbricate thrust sequence of the LongmenShan fault zone, where older faults have been rotated in the hanging walls of younger, listricfaults that rise from a gently dipping brittle-ductile transition zone. Eastward, the base of theschizosphere, or brittle-ductile transition zone, merges upward with subhorizontal decollementbeneath the Sichuan basin and finally ramps up in the Longquan Shan east of Chengdu (Chenet al. 2005, Jia et al. 2006, Burchfiel et al. 2008, Hubbard & Shaw 2009). As mentioned above,deformation within the Longmen Shan began during Late Triassic and Jurassic time and con-tinued to early Cenozoic time (Burchfiel et al. 1995, 2008; Wang & Meng 2008). Late Ceno-zoic deformation as manifested by the Wenchuan earthquake has been proposed to resultfrom eastward growth of the Tibetan Plateau as a consequence of lower crustal flow (Roydenet al. 1997, 2008; Clark & Royden 2000; Kirby et al. 2002; Clark et al. 2005). Thus, geody-namic processes of late Cenozoic and present-day deformation appear to differ from those ofMesozoic and early Cenozoic deformation. It is possible that the style of late Cenozoic deforma-tion has changed to high-angle, listric-reverse faulting with variable components of right-lateralstrike slip and crustal shortening. In any case, low-angle thrust faulting is not the style of de-formation seen from the late Cenozoic to the present. The 2008 Wenchuan earthquake andperhaps the abrupt topographic front of the Longmen Shan are manifestations of this new style ofdeformation.

The low-angle nature of the Longmen Shan faults revealed in seismic reflection profiling doesnot match the various observations associated with the Wenchuan earthquake mentioned above,all of which suggest a 70 to 80◦ dip angle above a depth of at least 10 km. As we mention above,aftershock hypocenter distributions appear to require that seismogenic faults dip steeply (morethan 70◦ to the northwest) above ∼15 km, become gently dipping (30 to 40◦) only below ∼15 km,and finally root into subhorizontal basal schizosphere (brittle-ductile transition zone) below 20 to22 km (Figure 6). Regional seismicity from 1970 to 2008 in the western Sichuan and aftershocksof the Wenchuan earthquake (Figure 10) do not delineate an active low-angle thrust system as the1999 Chi-chi earthquake aftershocks did in central Taiwan (Carena et al. 2002, Chang et al. 2007).Rather, microearthquakes seem to occur within the entire upper and middle crust (schizosphere)between depths of 10 and 25 km (Figure 10).

Fault-plane solutions of mainshock indicate slip on a fault dipping 30◦ to 40◦ (CENC 2008;Ji 2008, Nishimura & Yagi 2008, Y. Zhang et al. 2008, Global CMT Project). The preciselyrelocated epicenter of mainshock is only 8 km southwest of the primary surface rupture, theYingxiu-Shuiguan rupture on the Yingxiu-Beichuan fault. If the earthquake rupture took placeon a planar fault dipping 30 or 40◦, for the given mainshock hypocenter location the surfacerupture should appear 22–32 km east of the Yingxiu-Shuiguan rupture and within the Sichuanbasin. Together, the relocated hypocenter and the fault-plane solution of mainshock require slipon a high-angle, listric-reverse fault to produce the mapped surface ruptures along the Yingxiu-Beichuan and Guanxian-Jiangyou faults.

We propose that the structure responsible for the Wenchuan earthquake consists of imbricate,oblique, high-angle, listric-reverse faults. The faults dip ∼70◦ above 15-km depth, then dip 30 to

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 10Comparison of seismic pattern of the western Sichuan with middle Taiwan and Southern California. (a) Background seismicity (bluedots) of the western Sichuan region from 1970 to 2008; aftershocks associated with the 2008 Wenchuan earthquake (red crosses).(b) Cross section of earthquake distribution for the area indicated in (a).

370 Zhang et al.


>7.0

6.0–6.9

5.0–5.9

4.0–4.9

3.0–3.9

>8.0

5.0–5.9

4.0–4.9

3.0–3.9

2.0–2.9

FaultEarthquake magnitudeJanuary 1970–2008 Aftershock magnitude

99°E 102°E 105°E 108°E28°N

30°N

32°N

34°N

b

0

10

20

30

40

50–500 –400 –300 –200 –100 0 100 200

5000

Distance (km)

De

pth

(k

m)

Yingxiu-Beichuan faultA A'130°

A

A'A'

A

A'

a

Ele

va

tio

n (

m)



40◦ below ∼15-km depth, and presumably root into the subhorizontal brittle-ductile transitionzone below a depth of 20 to 22 km. Under the stress regime of pure shear, slip on a fault dipping30–40◦ is easy, according to various fracture criteria. The initial slip on the gentle dipping faultprobably caused the Coulomb stress changes that may in turn have triggered significant slip on thehigh-angle dipping fault above it to form the Wenchuan earthquake. The earthquake occurredon multisegmented ruptures, each of which is characterized by a different, complex geometry.The distribution of slip during the Wenchuan earthquake rupture presumably characterizes thestyle of late Cenozoic tectonic deformation of the Longmen Shan and eastern Tibet. It alsohelps explain the presence of a steep mountain relief (>4 km) with negligible coeval forelandsubsidence.

THE WENCHUAN EARTHQUAKE RUPTURE AND PRESEISMICACCUMULATION OF STRAIN

The 2008 Wenchuan earthquake is a consequence of interactions among multiple geologicalunits under a tectonic background in which the eastward growth of the Tibetan Plateau has beenimpeded by the tectonically stable Sichuan basin (Burchfiel et al. 1995, 2008; Royden et al. 1997;Clark & Royden 2000; Clark et al. 2005). The rheologically “soft” material in the middle andlower crust of eastern Tibet (Royden et al. 1997, 2008; Clark & Royden 2000; Clark et al. 2005;Liu et al. 2009) has been thickened, while the brittle upper crust has been obliquely pushed againstthe effectively rigid Sichuan basin on a high-angle reverse contact, the Longmen Shan fault zone.

Unlike an earthquake resulting from strain accumulation along a single fault, the Wenchuanearthquake involved three geological units: eastern Tibet, the Longmen Shan, and the Sichuanbasin (P.Z. Zhang et al. 2009) (Figure 11). Interactions among them caused strain accumulationin the Longmen Shan, and the strain was eventually released to form during the devastatingWenchuan earthquake. The three units behave differently during both interseismic and coseismicperiods.

The entire crust of eastern Tibet appears to be relatively weak (Wang et al. 2007; Yao et al.2008; L. Xu et al. 2007; Liu et al. 2009) and thus serves as a deforming unit. During interseismicperiods, significant deformation occurs mainly in eastern Tibet by convergence perpendicularto the Longmen Shan, right-lateral shear, and vertical movement (Figure 11a). Thus easternTibet functions as a strain-energy conveyor belt that continuously transfers its deformation intoaccumulating stress on the Longmen Shan fault zone.

The Longmen Shan thrust sheet consists of Precambrian metamorphosed crystalline basementwith relatively high strength such as the Baoxing, Pengguan, and Jiaoziding massifs (Burchfiel et al.2008, Wang & Meng 2008, Z. Xu et al. 2008). These massifs are oriented in a direction unfavorablefor slip; shortening normal to the strike of a high-angle reverse fault should increase its frictionalresistance to prohibit slip. The region around the fault thus serves as a stress-accumulation unit(Figure 11a) (P.Z. Zhang et al. 2009). The imbricate, listric, high-angle Longmen Shan faultzone remains locked during interseismic periods and accumulates stress at a low rate before anearthquake, so that preearthquake slip also accumulates slowly. During coseismic time, when theaccumulated stress exceeds the critical strength of the Longmen Shan fault zone, an earthquakeoccurs to release huge amounts of strain energy that have been slowly stored during the severalthousand–year interseismic period. The coseismic deformation occurs mostly in the interseismi-cally locked Longmen Shan (Figure 11b).

The Sichuan basin has behaved as a stable geological block since the late Mesozoic. It has amechanically strong lower crust and upper-mantle structure (Wang et al. 2007, Liu et al. 2009) and

372 Zhang et al.


Brittle-ductile transitionBrittle-ductile transition

Lon

gm

en S

hanEastern Tibet

Sichuanbasin

Eastern TibetSichuanbasin

0 100 200 300 400 500 600 700–2

0

2

4

6

8

10

12

14

Distance from eastern Tibet to Sichuan basin (km)

Eastern TibetSichuanbasin

Strain accumulation unit Strain release unit

102˚E 103˚E 104˚E 105˚E

31˚N

32˚N

33˚N

ChengduChengdu

DujiangyanDujiangyan

LixianLixian

MaerkangMaerkang

HeishuiHeishui

MianzhuMianzhu

AnxianAnxianBeichuanBeichuan

JiangyouJiangyouMaoxianMaoxian

PingwuPingwu

AbaAba

Chengdu

Dujiangyan

Lixian

Maerkang

Heishui

Mianzhu

AnxianBeichuan

JiangyouMaoxian

Pingwu

Aba

1 mm year–1

a b

c d

Vertical velocity

Town or villageBase station

Ve

loci

ty r

ela

tiv

e t

o t

he

Hu

an

na

n b

lock

(m

m y

ea

r–1)

Lon

gm

en

Sh

an

Lon

gm

en S

han

Right-lateral shear

Verticalthickening

Verticaldisplacement

Right-lateralstrike-slip

Crustal shortening

Horizontalshortening

Major active strand of the Longmen Shan fault

Major inactive strand of the Longmen Shan fault

Surface rupture associated with the Wenchuan earthquake

FaultSurface rupture

Velocity components perpendicular to the Longmen Shan fault zone

Right-lateral components parallel to the Longmen Shan fault

Figure 11Preearthquake geodetic deformation and postulated cartoon model of the Wenchuan earthquake occurrence. (a) Diagram illustratingconfiguration of preearthquake deformation. Eastern Tibet has been shortened, wrenched, and uplifted to accumulate strain, while theLongmen Shan has been locked to accumulate stress. (b) Diagram showing coseismic configuration. The stress accumulated during theinterseismic interval in the Longmen Shan is released through the Wenchuan earthquake rupture. (c) Preearthquake GPS velocityprofile from eastern Tibet to the Sichuan basin across the Longmen Shan fault zone. (d ) Vertical movement observed by levelingsurveys along lines from Aba to Chengdu and from Maoxian to Mianzhu. The first survey was taken in 1976, and the last was in 1997.The uncertainty associated with the rates should be regarded to larger than ± 1 mm year−1.

thus acts as a supporting unit to resist eastward movement of both eastern Tibet and the LongmenShan (Figure 11a). The supporting unit is a necessary condition for stress accumulation in theLongmen Shan fault zone, although minor deformation locally occurs along the western edge ofthe Sichuan basin.

Given the unique aspects of the Wenchuan earthquake—namely its size and occurrence in aregion of modest displacement rates—it is important to understand how the preseismic accumula-tion of strain occurred in the locking unit (the Longmen Shan). Figure 11c shows preearthquake



GPS velocity profiles with respect to the Sichuan basin. It is evident that no more than 2–3 mm year−1 of relative motion occurs between the Longmen Shan and the Sichuan basin across theLongmen Shan fault (King et al. 1997, Shen et al. 2005, Gan et al. 2007, P.Z. Zhang 2008),which is consistent with geological slip rates of less than 2 mm year−1 (Densmore et al. 2007,Zhou et al. 2007). Instead, linear gradients of both components of velocity across easternTibet manifested themselves as continuous compressive strain and right-lateral shearing beforethe earthquake with rates of 4–5 and 7–9 mm year−1, respectively, but distributed across a zoneat least 500 km wide (Figure 11c,d ). An important question is what fraction of this geodeticallymeasured accumulating strain in eastern Tibet is converted to seismogenic loading on the lockingunit (the Longmen Shan). In other words, is the strain in eastern Tibet permanent, or could it beelastic?

Suppose that deformation rates (8.1–10.3 mm year−1) in eastern Tibet were elastic and loadedthe Longmen Shan. In that case, the slip rate of the Longmen Shan fault zone would be significantlylarger than the rate of ∼2 mm year−1 obtained by studies of active faulting. The average coseismicdisplacement of ∼5 m (X. Xu et al. 2009) and a slip rate of ∼2 mm year−1 suggest a recurrenceinterval of ∼2500 years, which agrees with preliminary paleoseismic studies (P.Z. Zhang et al. 2008,Ran et al. 2008). Thus, the observations of fault slip rate and paleoseismology of the LongmenShan fault zone imply that a large portion of strain in eastern Tibet must be permanent. Onthe other hand, background seismicity before the Wenchuan earthquake (Figure 10) showing arelatively dense distribution in eastern Tibet and almost a vacancy in the Longmen Shan furthersuggests that a large portion of preseismic strain is accommodated by permanent deformation orby energy released seismically on faults in eastern Tibet.

Coseismic deformation also provides a test of whether the strain in eastern Tibet is permanentor elastic because elastic strain would imply significant coseismic deformation and strain release.Coseismic deformation associated with the Wenchuan earthquake, in fact, takes place mostlywithin the interseismically locked Longmen Shan fault zone. As shown in Figure 7, geodeticallymeasured regional coseismic deformation occurs mainly in the Longmen Shan and attenuatesrapidly toward both eastern Tibet and the Sichuan basin. This is also the case for the deformationfield obtained by InSAR (Sun et al. 2008, Shan et al. 2009). Aftershocks are confined within theLongmen Shan fault zone, along which the high intensities of X and XI on the China SeismicIntensity Scale are narrowly distributed, and decay rapidly away from it (Figure 12a). Strongground-motion data (X. Li et al. 2008) show that most of the stations that recorded peak groundacceleration in excess of 400 gal are located within the Longmen Shan region (Figure 12b).Therefore, the eastward growth of the Tibetan Plateau probably causes significant permanentdeformation in eastern Tibet, and only a small fraction of strain is loaded slowly onto the LongmenShan fault zone. The combination of interseismic locking, relatively high strength, and slowloading in the Longmen Shan fault produces a great earthquake with a long recurrent intervaloccurring in a region of slow strain accumulation. The 2008 Wenchuan earthquake is an exampleof such an event and is likely to characterize the behavior of earthquakes that rupture the LongmenShan fault zone.

−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→Figure 12Seismic intensity and strong ground motion of the 2008 Wenchuan earthquake. (a) Map showing that high seismic intensity andaftershocks are confined with the Longmen Shan fault zone. (b) Distribution of components of Peak Ground Acceleration (PGA) acrossthe Longmen Shan fault zone.

374 Zhang et al.


0 100 200 300 400 500 600–1000

–800

–600

–400

–200

0

200

400

600

800

1000

East-West componentsNorth-South componentsVertical components

Distance across the Longmen Shan fault zone from NW to SE (km)

Co

mp

on

en

ts o

f P

GA

(g

al)

b

Lon

gm

en

Sh

an

Sichuan basin and South China blockEastern Tibet

ChengduChengdu

Aba Aba

Songpan Songpan

Pingwu Pingwu

Nanchong Nanchong

Santai Santai

Mianyang Mianyang

Guanxian Guanxian

Wench

uan-Maoxia

n fault

Wench

uan-Maoxia

n fault

Yingxiu-Beichuan faultYingxiu-Beichuan fault

Guanxian-Jiangyou fa

ult


ultX

XIXI

IXIX

VIIVII

VIIIVIII

VIVI

Chengdu

Aba

Songpan

Pingwu

Nanchong

Santai

Mianyang

Guanxian

Magnitude

Other active fault

Wench

uan-Maoxia

n fault

Yingxiu-Beichuan fault


ultX

XI

IX

VII

VIII

VI

104°E 105°E 106°E 107°E103°E102°E

31°N

32°N

a

33°N

8.0–8.9

7.0–7.9

6.0–6.9

Town or village

Aftershock epicenters

Surface rupture

Intensity grade VIIntensity grade VIIIntensity grade VIIIIntensity grade IXIntensity grade XIntensity grade XI



SUMMARY POINTS

1. Post-earthquake geological, seismological, and geodetic investigations all point to a cas-cading rupture of four segments, each of which has different geometry and behavior. Thegeometric complexity indicates structural discontinuities or barriers along the seismo-genic fault that require complex rupture processes to overcome them.

2. Significant coseismic slip associated with the mainshock may occur in the shallow crustabove 10-km depth, as corroborated by the aftershock hypocenter distribution and geode-tic inversion for slip. This shallow slip then contributes to significant ground shaking.

3. The structure responsible for the Wenchuan earthquake is an imbricate, oblique, high-angle, listric, reverse fault that dips ∼70◦ above 15-km depth, then dips 30 to 40◦ below∼15-km depth, and finally roots into the subhorizontal brittle-ductile transition zonebelow a depth of 20 to 22 km. Although many believe that dip-slip faults become listricat depth and root into a ductile lower crust, the Wenchuan earthquake rupture offersperhaps the clearest demonstration that such a listric structure exists.

4. Unlike an earthquake resulting from strain accumulation along a single fault, theWenchuan earthquake involved three geological units: eastern Tibet, the Longmen Shan,and the Sichuan basin. Interactions among them caused slow strain accumulation in theLongmen Shan so that measurable preearthquake slip was minor. Coseismic deforma-tion, however, took place mostly within the interseismically locked Longmen Shan faultzone. The earthquake may have initiated from slip on a fault plane dipping 30–40◦ north-west in a depth range of 15 to 20 km beneath the southwestern segment of the rupturezone and may have triggered oblique slip on the high-angle faults at depths shallowerthan 15 km to produce the great Wenchuan earthquake.

FUTURE ISSUES

1. Although we have inferred the seismogenic structure of the Longmen Shan fault to be animbricate, high-angle, listric-reverse fault system, direct evidence is still needed throughhigh-resolution, deep-seismic-sounding profiling.

2. Post-earthquake deformation will provide important information on the physical prop-erties of both eastern Tibet’s lithosphere, and especially its lower crust, and the LongmenShan fault zone. Continuous GPS stations have been deployed after the earthquake. Anal-ysis of postearthquake relaxation should reveal valuable information on the behavior ofthe lower crust in general.

3. Up to now, strain processes associated with the Wenchuan earthquake have been givenin a qualitative, descriptive manner. To fully understand the interseismic, coseismic, andpostearthquake processes, quantitative experiments of the various relevant processes areneeded.

DISCLOSURE STATEMENT

The authors are not aware of any affiliations, memberships, funding, or financial holdings thatmight be perceived as affecting the objectivity of this review.

376 Zhang et al.


ACKNOWLEDGMENTS

This work has been a joint effort by many scientists who participated in emergency recovery andcomprehensive studies after the Wenchuan earthquake. We thank all participants for their valuablecontributions. We are grateful to the Sichuan Seismological Bureau for logistic support during thepostearthquake comprehensive studies. Thanks also go to Li Ming and Tian Liu from the ChinaSeismological Administration and Wang Li, Qi Guili, and Deng Yiwei from the Sichuan Seis-mological Bureau for their excellent administrative and logistical work that assured success of thecomprehensive postearthquake studies. We thank Peter Molnar, Leigh Royden, Clark Burchfiel,Rob van der Hilst, and Eric Kirby for comments and encouragement. The work is supported byfunds from the National Key Basic Research Program of China (2004CB418400), the State KeyLaboratory of Earthquake Dynamics (LED2008A01), the China Earthquake Administration, andthe National Science Foundation of China (40841013).

LITERATURE CITED

Ammon CJ, Ji C, Thio H-K, Robinson D, Ni S, et al. 2005. Rupture process of the 2004 Sumatra-Andamanearthquake. Science 308:1133–39

Avouac JP. 2003. Mountain building, erosion, and the seismic cycle in the Nepal Himalaya. Adv. Geophys.Vol. 46. 80 pp.

Bilham R, Gaur VK, Molnar P. 2001. Himalayan seismic hazard. Science 293:1442–44Bollinger L, Avouac JP, Cattin T, Pandey MR. 2004. Stress buildup in the Himalaya. J. Geophys. Res.

109:B11405Burchfiel BC, Chen Z, Liu Y, Royden LH. 1995. Tectonics of the Longmen Shan and adjacent regions,

Central China. Int. Geol. Rev. 37:661–735Burchfiel BC, Royden LH, van der Hilst RD, Hager BH, Chen Z, et al. 2008. A geological and geophysical

context for the Wenchuan earthquake of 12 May 2008, Sichuan, People’s Republic of China. GSA Today18(7):4–11

Carena S, Suppe J, Kao H. 2002. The active detachment of Taiwan illuminated by small earthquakes and itscontrol of first-order topography. Geology 30:935–38

CENC (China Earthquake Networks Center). 2008. http://www.csi.ac.cn/sichuan/index080512001.htmChang C-H, Wu Y-M, Zhao L, Wu T-T. 2007. Aftershocks of the 1999 Chi-Chi, Taiwan, earthquake: the

first hour. Bull. Seismol. Soc. Am. 97:1245–58Chen J, Liu Q, Li S, Guo B, Li Y, et al. 2009. Seismotectonics study by relocation of the Wenchuan Ms8.0

Earthquake sequence. Chin. J. Geophys. 52:390–97 (In Chinese with an English abstract)Chen SF, Wilson CJL. 1996. Emplacement of the Longmen Shan Thrust-Nappe Belt along the eastern

margin of the Tibetan Plateau. J. Struct. Geol. 18:413–30Chen Z, Burchfiel BC, Liu Y, King RW, Royden LH, et al. 2000. Global Positioning System measurements

from eastern Tibet and their implications for India/Eurasia intercontinental deformation. J. Geophys. Res.105:16215–27

Chen Zh, Jia D, Zhang Q, Wei G, Li B, et al. 2005. Balanced cross-section analysis of the fold-thrust belt ofthe Longmen mountains. Acta Geol. Sinica 16:404–12

Clark MK, Bush J, Royden LH. 2005. Dynamic topography produced by lower crustal flow against rheologicalstrength heterogeneities bordering the Tibetan Plateau. Geophys. J. Int. 162:575–90

Clark MK, Royden LH. 2000. Topographic ooze: building the eastern margin of Tibet by lower crustal flow.Geology 28:703–6

Densmore AL, Ellis MA, Li Y, Zhou R, Hancock GS, Richardson N. 2007. Active tectonics of the Beichuanand Pengguan faults at the eastern margin of the Tibetan Plateau. Tectonics 26:TC4005

Dietz LD, Ellsworth WL. 1997. Aftershocks of the Loma Prieta earthquake and their tectonic implications.U.S. Geol. Surv. Prof. Pap. 150-D:5–48

Global Centroid-Moment-Tensor (CMT) Project, http://www.globalcmt.org/



Hartzell SH, Heaton TH. 1983. Inversion of strong ground motion and teleseismic waveform data for the faultrupture history of the 1979 Imperial Valley, California, earthquake. Bull. Seismol. Soc. Am. 73(6A):1553–83

Hatzfeld D, Karakostas V, Ziazia M, Selvaggi G, Leborgne S, et al. 1997. The Kozani-Grevena (Greece)earthquake of May 13, 1995 revisited from a detailed seismological study. Bull. Seismol. Soc. Am. 87:463–73

Hauksson E, Jones L, Hutton K. 2002. The 1999 Mw 7.1 Hector Mine, California, earthquake sequence:complex conjugate strike-slip faulting. Bull. Seismol. Soc. Am. 92:1154–70

Hauksson E, Shearer P. 2005. Southern California hypocenter relocation with waveform cross-correlation,part 1: results using the double-difference method. Bull. Seismol. Soc. Am. 95:896–903

Hubbard J, Shaw JH. 2009. Uplift of the Longmen Shan and Tibetan Plateau, and the 2008 Wenchuan(M = 7.9) earthquake. Nature 458:194–97

Gan W, Zhang P-Z, Shen Z-K, Niu Z, Wang M, et al. 2007. Present-day crustal motion within the TibetanPlateau inferred from GPS measurements. J. Geophys. Res. 112:B08416

Ji C. 2008. Preliminary result of the May 12, 2008 Mw 7.97 ShiChuan earthquake. http://www.geol.ucsb.edu/faculty/ji/big_earthquakes/2008/05/12/ShiChuan.html

Jia D, Wei G, Chen Z, Li B, Zeng Q, et al. 2006. Longmen Shan fold-thrust belt and its relation to the westernSichuan Basin in central China: new insights from hydrocarbon exploration. AAPG Bull. 90:1425–47

Kanamori H, Anderson DL. 1975. Theoretical basis of some empirical relations in seismology. Bull. Seismol.Soc. Am. 65:1073–95

King RW, Shen F, Burchfiel BC, Royden LH, Wang E, et al. 1997. Geodetic measurement of crustal motionin southwest China. Geology 25:179–82

Kirby E, Reiners PW, Krol MA, Whipple KX, Hodges KV, et al. 2002. Late Cenozoic evolution of the easternmargin of the Tibetan Plateau: inferences from 40Ar/39Ar and (U-Th)/He thermochronology. Tectonics21(1):1001

Kirby E, Whipple K, Harkins N. 2008. Topography reveals seismic hazard. Nat. Geosci. 1:485–87Kumar S, Wesnousky SG, Rockwell TK, Briggs RW, Thakur VC, Jayangondaperumal R. 2006. Paleoseismic

evidence of great surface rupture earthquakes along the Indian Himalaya. J. Geophys. Res. 111:B03304Kumar S, Wesnousky SG, Rockwell TK, Ragona D, Thakur VC, et al. 2001. Earthquake recurrence and

rupture dynamics of Himalayan frontal thrust, India. Science 294:2328–31Lave J, Avouac J-P. 2000. Active folding of fluvial terraces across the Siwaliks Hills, Himalayas of central

Nepal. J. Geophys. Res. 105:5735–70Lave J, Yule D, Sapkota S, Basant K, Madden C, et al. 2005. Evidence for a great medieval earthquake

(∼1100 A.D.) in the Central Himalayas, Nepal. Science 307:1302–5Lay T, Kanamori H, Ammon CJ, Nettles M, Ward SN, et al. 2005. The great Sumatra-Andaman earthquake

of 26 December 2004. Science 308(5725):1127–33Li Ch, Zhang PZ, Yin J, Min W. 2009. Late Quaternary left-lateral slip rate of the Haiyuan fault, northeastern

margin of the Tibetan Plateau. Tectonics 28:TC5010Li H, Fu X, Van Der Woerd J. 2008. Surface rupture associated with the Wenchuan earthquake and its oblique

slip. Acta Geol. Sinica 82:1623–43Li SL, Lai XL, Yao ZX, Yang Q. 2009. Study on fault zone structures of northern and southern portions of

the main central fault generated by Ms = 8.0 Wenchuan earthquake using fault-zone trapped waves.Acta Seismol. Sinica In press

Li SL, Zhang XK, Fan JC. 2005. Study on rupture zone of the M = 8.1 Kunlun Mountain earthquake usingfault-zone trapped waves. Acta Seismol. Sinica. 27:43–52

Li X, Zhou Z, Huang M, Wen R, Yu H, et al. 2008. Preliminary analysis of strong-motion recordings fromthe magnitude 8.0 Wenchuan, China, earthquake of 12 May 2008. Seismol. Res. Lett. 79:844–54

Li YG, Aki K, Admas D, Hasemi A, Lee WHK. 1994. Seismic guided waves trapped in the fault zone of theLanders, California, earthquake of 1992. J. Geophys Res. 99:11705–22

Li YG, Chen P, Cochran ES, Vidale JE, Burdette T. 2006. Seismic evidence for rock damage and healingon the San Andreas fault associated with the 2004 M 6.0 Parkfield Earthquake. Bull. Seismol. Soc. Am.96:S349–63

Li YG, Leary PG. 1990. Fault zone trapped seismic waves. Bull. Seismol. Soc. Am. 80:1245–71

378 Zhang et al.


Li YG, Vidale JE, Day SM, Oglesby DD, and the SCEC Field Working Team. Study of the. 1999. M 7.1Hector Mine, California, earthquake fault plane by trapped waves. 2002. Bull. Seismol. Soc. Am. 92:1318–32

Liu J, Zhang Zh, Wen L. 2008. The Ms 8.0 Wenchuan earthquake co-seismic rupture and its tectonicimplications—an out-of-sequence thrusting event with slip partitioned on multiple faults. Acta Geol.Sinica 82:1707–22 (In Chinese with an English abstract)

Liu Q, Li Y, Chen J, Guo B, et al. 2009. Wenchuan Ms8.0 earthquake: Preliminary study of the S-wave velocitystructure of the crust and upper mantle. Chin. J. Geophys. 52:309–319 (In Chinese with an English abstract)

McCaffrey R. 2009. The tectonic framework of the Sumatran subduction zone. Annu. Rev. Earth Planet. Sci.37:345–66

Mendoza C, Hartzell SH. 1988. Aftershock patterns and mainshock faulting. Bull. Seismol. Soc. Am. 78:1438–49Nishimura N, Yagi Y. 2008. Rupture process for May 12, 2008 Sichuan earthquake (preliminary result).

http://www.geol.tsukuba.ac.jp/∼nisimura/20080512/Ouyed M, Yielding G, Hatzfeld D, King GCP. 1983. An aftershock study of the El Asnam (Algeria) earthquake

of 1980 October 10. Geophys. J. R. Astron. Soc. 73:605–39Ran Y, Chen L, Chen G, Yin J, Chen J, et al. 2008. Primary analysis of in-situ recurrence of large earthquake

along seismogenic fault of the Ms 8.0 Wenchuan Earthquake, China. Seismol. Geol. 30:630–43 (In Chinesewith English abstract)

Ran Y, Shi X, Wang H, Chen L, Chen J, et al. 2010. The maximum vertical co-seismic displacement andthe style of deformation associated with the 2008 Wenchuan earthquake. Chin. Sci. Bull. 54:In press (InChinese)

Royden LH, Burchfiel BC, King RW, Wang E, Chen Z, et al. 1997. Surface deformation and lower crustalflow in eastern Tibet. Science 276:788–90

Royden L, Burchfiel BC, van der Hilst RD. 2008. The geological evolution of the Tibetan Plateau. Science321:1054–58

Shan X, Qu C, Song X, Zhang G, Liu Y, et al. 2009. Coseismic surface deformation caused by the Wenchuanearthquake from InSAR data analysis. Chin. J. Geophys. 52:496–504 (In Chinese with an English abstract)

Shen ZK, Lu J, Wang M, Burgmann R. 2005. Contemporary crustal deformation around the southeastborderland of the Tibetan Plateau. J. Geophys. Res. 110:B11409

Shen ZK, Sun J, Zhang PZ, Wan Y, Wang M, et al. 2009. Slip maxima at fault junctions and rupturing ofbarriers during the 2008 Wenchuan earthquake. Nat. Geosci. 2:718–24

Sibson RH. 1986. Earthquakes and rock deformation in crustal fault zones. Annu. Rev. Earth Planet. Sci.14:149–75

Sibson RH. 1989. Earthquake faulting as a structural process. J. Struct. Geol. 11:1–14Sieh K, Jones L, Hauksson E, Hudnut K, Eberhart-Phillips D, et al. 1993. Near-field investigations of the

Landers earthquake sequence, April to July 1992. Science 260:171–76Sun J, Liang F, Shen ZK, Xu X. 2008. InSAR deformation observation and preliminary analysis of the Ms 8.0

Wenchuan earthquake. Seismol. Geol. 30:489–95 (In Chinese with an English abstract)Thurber C, Zhang H, Waldhauser F, Hardebeck J, Michael A, Eberhart-Phillips D. 2006. Three-dimensional

compressional wavespeed model, earthquake relocations, and focal mechanisms for the Parkfield,California, region. Bull. Seismol. Soc. Am. 96:S38–49

Waldhauser F, Ellsworth WL. 2000. A double-difference earthquake location algorithm: method and appli-cation to the Northern Hayward Fault, California. Bull. Seismol. Soc. Am. 90:1353–68

Wang CY, Han WB, Wu JP, Lou H, Chan WW. 2007. Crustal structure beneath the eastern margin of theTibetan plateau and its tectonic implications. J Geophys. Res. 112:B07307

Wang CY, Mooney WD, Ding ZF, Yang JS, Yao ZX, Lou H. 2009. Shallow seismic structure of Kunlun faultzone in northern Tibetan Plateau, China: implications for the 2001 Ms8.1 Kunlun earthquake. Geophys.J. Int. 177(3):978–1000

Wang E, Meng Q. 2008. Discussion on tectonic evolution of the Longmen Shan fault zone. Sci. China Ser. D38:1221–33

Wang W, Zhao L, Li J, Yao Zh. 2008. Rupture process of the Ms 8.0 Wenchuan earthquake of Sichuan,China. Chin. J. Geophys. 51:1403–10 (In Chinese with an English abstract)



Working Group of the Crustal Motion Observation Network of China Project. 2008. Coseismic displacementfield of the 2008 Ms 8.0 Wenchuan earthquake determined by GPS. Sci. China Ser. D 38:1195–206 (InChinese)

Xu L, Rondenay S, Van Der Hilst RD. 2007. Structure of the crust beneath the southeastern Tibetan Plateaufrom teleseismic receiver functions. Phys. Earth Planet. Inter. 165:176–93

Xu X, Wen X, Ye J, Ma B, Chen J, et al. 2008. The Ms 8.0 Wenchuan earthquake surface ruptures and itsseismogenic structure. Seismol. Geol. 30:597–629 (In Chinese with an English abstract)

Xu X, Wen X, Yu G, Chen G, Klinger Y, et al. 2009. Coseismic reverse- and oblique-slip surface faultinggenerated by the 2008 Mw 7.9 Wenchuan earthquake, China. Geology 37:515–18

Xu Z. 2009. Wenchuan Fault Scientific Drilling Program (WFSD). http://www.wfsd.org/Xu Z, Ji S, Li H, Hou L, Fu X, Cai Z. 2008. Uplift of the Longmen Shan range and the Wenchuan earthquake.

Episodes 31:291–301Yao H, Beghein C, van der Hilst RD. 2008. Surface-wave array tomography in SE Tibet from ambient seismic

noise and two-station analysis, II. Crustal and upper-mantle structure. Geophys. J. Int. 173:205–19Yeats RS, Sieh K, Allen CR. 1997. The Geology of Earthquakes. Oxford Univ. Press. 576 pp.Zhang PZ, Molnar P, Burchfiel BC, Royden L, Zhang W, et al. 1988. Bounds on the Holocene slip rate along

the Haiyuan fault, north-central China. Quat. Res. 30:151–64Zhang PZ, Slemmons DB, Mao F. 1991. Geometric pattern, rupture termination, and fault segmentation of

the Dixie Valley–Pleasant Valley active normal fault system. J. Struct. Geol. 12:165–76Zhang PZ, Deng Q, Zhang G, Ma J, Gan W, et al. 2003. Active tectonic blocks and strong earthquakes in

continental China. Sci. China Ser. D 33(Suppl.):13–24Zhang PZ, Shen Z, Wang M, Gan W, Burgmann R, et al. 2004. Continuous deformation of the Tibetan

Plateau from Global Positioning System data. Geology 32:809–12Zhang PZ, Xu XW, Wen XZ, Rang RK. 2008. Slip rates and recurrence intervals of the Longmen Shan active

fault zone and tectonic implications for the mechanism of the May 12 Wenchuan earthquake, 2008,Sichuan, China. Chin. J. Geophys. 51:1066–73 (In Chinese with an English abstract)

Zhang PZ. 2008. Tectonic deformation, strain partitioning, and crustal dynamics of the western Sichuanregion. Sci. China Ser. D 38:1041–56 (In Chinese)

Zhang PZ, Wen X, Xu X, Gan W, Wang M, et al. 2009. Model of strain accumulation and release associatedwith the 2008 Wenchuan, Sichuan, China earthquake. Chin. Sci. Bull. 54:944–53 (In Chinese)

Zhang Y, Feng WP, Xu LS, Zhou CH, Chen YT. 2008. Spatio-temporal rupture process of the 2008 greatWenchuan earthquake. Sci. China Ser. D 52:145–154

Zhang Y, Xu L, Chen Y. 2009. Spatial-temporal variation of the source mechanism of the 2008 great Wenchuanearthquake. Chin. J. Geophys. 52:379–89

Zheng Y, Ma H, Lu J, Ni S, Li Y, Wei S. 2009. Source mechanism of strong aftershocks (Ms ≥ 5.6) of the2008/05/12 Wenchuan earthquake and the implication for seismotectonics. Sci. China Ser. D 52(6):739–53

Zhou R, Li Y, Densmore AL, Ellis MA, He Y, et al. 2007. Active tectonics of the Longmen Shan region ofthe eastern margin of the Tibetan plateau. Acta Geol. Sinica 81:593–604

Zhu J. 2008. Background of the Wenchuan earthquake occurrence in deep structure and lithospheric dynamics.J. Chengdu Univ. Technol. 35:348–56 (In Chinese with an English abstract)

380 Zhang et al.

Oblique, High-Angle, Sichuan, Chinascec.ess.ucla.edu/~zshen/publ/Areps10Zhang.pdfOn May 12, 2008, a devastating earthquake struck densely populated Sichuan Province, China ... Sichuan

Documents